Equipment Testing and Validation ASSESSMENT OF FRACTIONAL EXPIRED GASES AND AIR FLOW BY AN AMBULATORY METABOLIC ANALYZER

Transcription

1 Ambulatory Metabolic Analysis 20 JEPonline Journal of Exercise Physiologyonline Official Journal of The American Society of Exercise Physiologists (ASEP) ISSN An International Electronic Journal Volume 2 Number 4 October 1999 Equipment Testing and Validation ASSESSMENT OF FRACTIONAL EXPIRED GASES AND AIR FLOW BY AN AMBULATORY METABOLIC ANALYZER PATRICK L. JACOBS, MARK S. NASH AND C. DAVID MINTZ Departments of Neurological Surgery and Orthopaedics & Rehabilitation, and The Miami Project University of Miami School of Medicine, Miami, Florida ABSTRACT ASSESSMENT OF FRACTIONAL EXPIRED GASES AND AIR FLOW BY AN AMBULATORY METABOLIC ANALYZER. PATRICK L. JACOBS, MARK S. NASH AND C. DAVID MINTZ. JEPONLINE, 1999, 2(4): Fractional expired oxygen (F E O 2 ) and carbon dioxide (F E CO 2 ), and air volume flow were compared using an ambulatory metabolic analyzer (KB1-C, Aerosport, Inc.) and a laboratory metabolic analyzer (Horizon MMC, Sensormedics, Inc.). Ten subjects with paraplegia exercised to exhaustion using their arms, and ten nondisabled subjects ran to exhaustion on a treadmill. Simultaneous F E O 2 and F E CO 2 measurements were taken during exercise by inserting the KB1-C pneumotachometer in series with the MMC. Measurement accuracy of air flow was examined by sending known air volumes through the KB1-C at three pneumotachometer flow settings (low = 3 to 30 L. min -1 ; medium = 3 to 120 L. min -1 ; and high = 3 to 200 L. min -1 ). Results showed a significant positive association for F E O 2 (R 2 = 0.94) with minimal but statistically significant absolute differences (SE E = 0.26%, p<0.001)). The R 2 for F E CO 2 was 0.89, although average KB1-C measurements were slightly lower (SE E = 0.28%, p<0.001). Differences from the reference standard for the low flow pneumotachometer were 12.3 and 15.5% at 3 L/min and 6 L/min, respectively, but averaged between 2.5 and 4.0% at 9-45 L/min. For the medium flow setting, average differences from the reference standard ranged from 0.5% at 72 L/min to 4.6% at 117 L/min. For the high flow setting, measurements varied dramatically from the reference standard at flow rates up to nine L/min, and by % at 27 to 54 L/min. Otherwise, differences of less than 2% were observed at rates of L/min. This study observed similar measurements for F E O 2 and F E CO 2 when measured by an ambulatory metabolic analyzer and a laboratory standard, and found that accurate measurement of air flow depended upon careful matching of the pneumotachometer setting with the expected air volume ranges. Key Words: Exercise, Metabolism, Ventilation, Fractional Gases, Portable INTRODUCTION Metabolic analysis using open-circuit spirometry is a widely used standard for assessing fitness, calculating work intensities, and estimating caloric expenditures. Unfortunately, the size, relative immobility, and need for residential/commercial (110 or 240 V) electrical current restricts the use of most metabolic analyzers to the controlled laboratory setting. Such restrictions also limit the ability of investigators to address study questions about physical activity when conducted under field conditions of ambient temperature and humidity. Widespread use and universal acceptance of portable

2 Ambulatory Metabolic Analysis 21 metabolic analyzers has been challenged by their size (1-3), lack of direct carbon dioxide (CO 2 ) gas measurement (4-7), and poor measurement validity 3,5,9,11). However, an ambulatory metabolic testing system claiming to solve these problems has recently been marketed (Aerosport KB1-C, Aerosport, Inc., Ann Arbor, MI). This system weighs about 1.0 kg, has dimensions that favor its portability, measures both oxygen (O 2 ) and CO 2 gases, and operates under rechargeable battery power. Unlike many laboratory analyzers, it uses a flat plate pressuredifferential pneumotachometer to measure expired gas volumes, which eliminates the need for bulky expired gas lines. In addition, the sampling of expired air from the pneumotachometer eliminates the needs for an expired gas-mixing chamber. Despite favorable physical characteristics and portability, a study comparing measurements obtained from the KB1-C unit with those of an industry standard has yet to be reported for both rest and exercise conditions. Various study methods have been used to make such comparisons for other portable metabolic analyzers. Some studies have compared VO 2 measurements at matched work intensities during separate exercise tests that use identical test protocols (1,4-7). Such analyses may be compromised by small differences in day-to-day exercise physiology and test-retest calibration errors. Other studies have simultaneously measured VO 2 during the same exercise test by placing the portable gas analyzer in series (2,3) with a laboratory standard. Unfortunately, the physics of air flow by which a pneumotachometer accurately measures ventilation volume may render such methods suspect. For example, accurate pneumotachometer measurement of minute ventilation (V E ) favors bidirectional air flow under low pressure conditions (11), while many open-circuit methods used for such comparisons employ unidirectional flow in a closed circuit of descending ventilation tube diameters and building pressure. The accuracy of in-series air flow comparisons may therefore be compromised, as the KB1-C computer program assumes the air pressure on the downstream side of its pneumotachometer plate (by calibration) to be ambient. The accurate determination of VO 2 and VCO 2 by open-circuit spirometry requires simultaneous measurement of fractional expired gasses (O 2 and CO 2 ) and minute ventilation. Several investigations have compared VO 2 alone, or VO 2 and V E together (2-6) during graded and steady state work. For those cases in which differences between portable and laboratory standards were observed, the source of this difference has never been established (e.g. measurement error for gas concentration versus volume). The isolation of such measurement error is important, as the KB1-C permits the selection of three fixed pneumotachometer settings. While each pneumotachometer may be considered accurate throughout a defined flow range, that specified range may be exceeded during exercise testing. Measurement of VO 2 alone, as reported by most studies, fails to identify the source of this potential error. Thus, the purpose of this study was twofold. First, the study compared simultaneous measurements of fractional expired oxygen (F E O 2 ) and carbon dioxide (F E CO 2 ) gas concentrations obtained during graded exercise on the KB1-C ambulatory metabolic analyzer with those of a laboratory standard. Second, we compared KB1-C pneumotachometer volume measurements obtained both within and outside its three rated ranges with those of a calibration standard. METHODS Instrumentation The KB1-C is a battery powered, microprocessorcontrolled metabolic analyzer designed for ambulatory and field-based studies. The unit computes VO 2, VCO 2, and respiratory exchange ratios using measurements derived from gas volume flow and fractional inspired and expired gas concentrations. Ventilation volumes are calculated using a flat-plate orifice within an open pneumotachometer. Air flow over the orifice produces a drop in pressure proportional to the square root of its flow. Pressure differentials across the plate are measured via two pneumotachometer

3 Ambulatory Metabolic Analysis 22 pressure lines using a bonded strain gauge positioned across the orifice plate. The pneumotachometer has three flow ranges (low, medium, and high). The low flow range (LFR) is recommended for flow rates between two and 30L/min, the medium flow range (MFR) from 10 to 120 L/min, and the high flow range (HFR) from 25 to 200 L/min. Fractional concentrations of expired O 2 and CO 2 are measured in gas samples drawn from the pneumotachometer. Oxygen concentration is assayed by a galvanic fuel cell contained in the unit, and computed using a line of identity drawn through null and the ambient air O 2 concentration (20.93%). Carbon dioxide concentration is measured using nondispersive infrared gas analysis, which uses ambient CO 2 as a reference. These O 2 and CO 2 values are then corrected for ambient temperature and pressure. The reference system used in this study was a Horizon System MMC (Sensormedics, Inc.). This system also measures O 2 and CO 2 concentrations using a galvanic fuel cell and nondispersive infrared gas analysis, respectively. Both analyzers were calibrated before each exercise test using certified gases of known concentration. Subjects Twenty male volunteers between the ages of years participated in this study. Ten of these subjects had neurologically complete paraplegia (ASIA A) between T 5 and T 12 for more than one year. The remaining ten subjects had no neurological deficits. All subjects were in good health - operationally defined as asymptomatic for acute treatable illnesses - and had no known musculoskeletal or cardiac conditions that would otherwise limit their exercise activities. Written consent to participate in this study was obtained according to guidelines of the University of Miami Medical Sciences Subcommittee for the Protection of Human Subjects. Exercise Testing All subjects underwent graded exercise testing to volitional fatigue. Criteria for exercise termination were consistent with those recommended by the American College of Sports Medicine (8). Subjects with paraplegia underwent continuous graded arm tests on an Upper Body Ergometer (Cybex, Inc., Owatonna, MN) starting at an initial workload of 400 kg/m/min and progressing by 100 kg/m/min every three minutes. Subjects without neurological deficits underwent graded continuous treadmill exercise using the protocol of Bruce (9). Gas and Volume Measurements Fractional expired gas percentages for O 2 and CO 2 during each exercise test were simultaneously measured using the KB1-C and the MMC. These systems were placed in series, with the KB1-C MFR pneumotachometer positioned immediately before the gas mixing chamber of the MMC. To ensure complete purging of calibration and ambient gases from the mixing chamber and gas sampling lines, resting gas collection was started four minutes before the start of exercise testing. Measurement of Gas Volumes The accuracy of gas flow measured by the three KB1-C pneumotachometers settings was assessed on another occasion by manually injecting ambient air through the pneumotachometers. A certified three liter syringe (Hans Rudolf Inc., Kansas City, MO) was used to pass known volumes of air through the KB1-C pneumotachometer within 20 second periods, the shortest time interval over which the system can be programmed to calculate and display information. The pneumotachometer was connected to the syringe by a six centimeter length of 3.5 centimeter diameter corrugated plastic tubing. Each of the three pneumotachometer range settings (LFR, MFR, and HFR) was tested at multiple flow rates across and beyond the ventilation ranges recommended by the manufacturer. The manufacturer recommended ranges for the LFR, MFR, and HFR are two to 30 L/min, 10 to 120 L/min, 25 to 200 L/min, respectively. The three pneumotachometer settings were evaluated at three, six, and nine L/min, and at increasing nine L/min intervals thereafter. The upper limits of the flow rate testing were 63, 126 and 216 L/min for the LFR, MFR, and HFR, respectively. Thus this protocol

4 Ambulatory Metabolic Analysis 23 allowed the examination of possible measurement inaccuracies for flow rates exceeding the recommendations for each setting. Ten ventilation tests were performed at each of these flow rates, making a grand total of 510 flow tests. Statistical Analysis Fractional expired gas concentrations were examined between KBC-1 and MMC by pooling the simultaneous measurements of the twenty subjects across workloads. Regression analyses were performed, and correlation coefficients calculated, which examined the associations between the simultaneously derived KB1-C and MMC measurements of F E O 2 and F E CO 2. Measurements derived from the two metabolic analyzers were compared using Student s paired t-tests. Standard errors of estimate (SE E ) were calculated using the formula: SE E = S y (1- r 2 ) ½, where r 2 is the reliability coefficient and S y is the standard deviation of the MMC values. Comparison of reference gas volumes with those obtained from the KB1-C was made using Student s paired t-tests. Means and standard deviations for KB1-C measurements were calculated for each flow rate. The measurements of air flow made by the KB1- C were compared with those of the standard across all tested flow rates for each of the three pneumotachometer settings (Student s paired t-tests and SE E ). Linear regression analysis was performed for each pneumotachometer range to examine the association between values obtained from the KB1-C and the reference system. Statistical significance for all tests was accepted at p<0.05. Based on 20 subjects, acceptable power of at least 0.80, and a standard deviation for ventilation of 0.25 L/min, the mean difference that could be detected was 0.20 L/min. Pooling the data by the ten trial repetitions and multiple pneumotachometer settings provided additional improvements to statistical power. RESULTS Comparisons of Reference and KB1-C Measurements for FEO 2 and F E CO 2 Figure 1 shows subpeak and peak F E O 2 values simultaneously measured by the MMC analyzer and the KB1-C. A significant positive association for this relationship was obtained (r 2 =0.94), although significant differences between the two means were also observed. These differences were explained by lower F E O 2 values measured by the KB1-C than the MMC (p<0.001). However, the magnitude of test differences was small, as the SE E for F E O 2 was only 0.26%. This was less than 1.7% of the mean reference values for F E O 2 and well within the sensitivity error for many metabolic analyzers (10). When examining the same relationships for F E CO 2, the coefficient of determination was 0.89, although the means measured by the KB1-C were lower than the MMC (p<0.001) (Figure 2). The SE E for F E CO 2 values was 0.28 %, which was 6.5% of the mean reference value. Figure 1 - The relationship between expired air O 2 concentration measured using the Sensormedics MMC and the Aerosport KB1-C during graded incremental exercise. Comparisons of Reference and KB1-C Measurements for Air Flow Measurements of air flow for all three pneumotachometer settings are shown in Table 1. These values represent the mean and standard deviation of ten tests at each selected flow volume. For the LFP, large differences from the reference

5 Ambulatory Metabolic Analysis 24 Table 1: Air volume measurements (mean±sd) and mean errors using three pneumotachometer settings of the Aerosport KB1-C ambulatory metabolic analyzer (ten tests each) in reference to volumes of air injected with a calibration syringe. High Flow Calibation Low Flow Medium Flow (L/min) mean ± SD mean error mean ± SD Mean error mean ± SD mean error ± 0.22 (12.3%) 1.89 ± 0.37 (37.0%) 0.15 ± 0.05 (95.0%) ± 0.08 (15.5%) 5.82 ± 0.15 (3.0%) 0.65 ± 0.29 (89.2%) ± 0.07 (4.0%) 8.84 ± 0.08 (1.8%) 2.87 ± 0.57 (68.1%) ± 0.07 (3.5%) ± 0.09 (2.9%) ± 0.44 (14.1%) ± 0.41 (2.8%) ± 0.55 (1.5%) ± 0.21 (8.8%) ± 0.27 (3.2%) ± 0.47 (2.6%) ± 0.23 (8.2%) ± 0.08 (2.5%) ± 0.78 (1.5%) ± 0.22 (7.5%) ± 3.71 (10.5%) ± 0.32 (0.9%) ± 0.41 (7.0%) ± 0.74 (19.3%) ± 0.27 (1.0%) ± 0.37 (5.6%) ± 0.21 (0.5%) ± 0.29 (3.9%) ± 0.24 (0.6%) ± 0.26 (3.2%) ± 0.60 (0.5%) ± 0.46 (2.2%) ± 0.41 (0.8%) ± 0.31 (0.5%) ± 1.10 (1.1%) ± 0.40 (1.2%) ± 1.51 (4.6%) ± 0.41 (1.2%) ± 1.49 (7.5%) ± 0.36 (1.6%) ± 0.60 (0.8%) ± 0.95 (1.5%) ± 0.34 (2.3%) ± 0.57 (0.8%) ± 0.56 (0.6%) ± 2.83 (1.3%) ± 1.68 (1.2%) ± 1.58 (1.8%) ± 1.58 (3.1%) ± 2.78 (3.3%) standard were observed at flow rates of 3 L/min and 6 L/min (12.3 and 15.5% different, respectively). Differences averaged between 2.5 and 4.0% for rates ranging from 9-45 L/min, and increased to errors of 10.5% and 19.3% at 54 and 63 L/min, respectively. The SE E of the LFP measurements across flow rates ranging from 3 to 45 L/min was 4% of the mean reference values. Figure 3a compares these measurements with a line of unity. The greatest source of variation within this flow

6 Ambulatory Metabolic Analysis 25 range occurred at the two lowest flow rates tested, 3 and 6 L/min. The SE E of the LFP flow rate measurements obtained across flow rates ranging from 9 to 45 L/min was 2.4% of the mean of the reference system values. The absolute differences in measurement across this range (9 to 45 L/min) ranged from 0.63 to 1.12 L/min. flow rate measurements differed dramatically from the reference system values at flow rates equal to, or less than, 9 L/min; and by about 7-8% at the 27 to 54 L/min rates. Otherwise, differences of less than 2% from the standard were observed at flow rates of L/min. At flow rates from 72 to 216 L/min, the SE E of the HFP measurements represented 1.5% of the mean reference values. Within this range, the absolute differences between the HFP and the reference standard ranged from 0.46 to 3.64 L/min. DISCUSSION Figure 2 - The relationship between expired air CO 2 concentration measured using the Sensormedics MMC and the Aerosport KB1-C during graded incremental exercise. Air flow volume measurements using the MFP are reported in Table 1. The largest deviations from the standard in this range (3 to 126 L/min) occurred at flow rates well below and above the recommended range for the MFP. Otherwise, average differences from the reference ranged from 0.5% at 72 L/min to 4.6% at 117 L/min. Figure 3b displays the relationship of these measurements when compared with a line of unity. The SE E for the MFP values at flow rates between 6 and 117 L/min was 2.9% of the mean reference values and the absolute differences ranged from 0.18 to 5.35 L/min. Mean values and standard deviations of the HFP flow volume assessments are shown in Table 1, while Figure 3c graphically presents the comparisons of the KB1-C measurements with the line of unity. The HFP The findings of this study investigating the accuracy of an ambulatory metabolic analyzer agree with those of previous reports showing reliable and valid measurements between selected portable systems and standard laboratory metabolic analysis models. This consistency appears to be dependent upon direct measurement of both O 2 and CO 2, as the accuracy of measurement for portable analyzers that algebraically computed VCO 2 from a measured VO 2 and an assumed RER of unity failed to be as accurate (4,5,7). Several studies have compared a portable analyzer having both O 2 and CO 2 sensors with laboratory standards. For the Teem 100 (11), VO 2 measured during graded work performed on two occasions showed no differences from VO 2 measured with a computer-based component system using a ventilation meter and gas analyzers during three steady-state exercise trials of walking, jogging, and running (1). Others have placed this portable analyzer in series with other laboratory analyzers. In one study, no significant differences of VO 2 were observed between the Teem 100 and a laboratory standard during incremental treadmill running and stepping (3). In another study where the analysis systems were placed in series, no differences were observed for VO 2 during incremental cycling (2). The present investigation was novel in several respects. This study was the first to assess the measurement accuracy of O 2 and CO 2 gas sensors and the expired air flow in an ambulatory metabolic analysis system. Previous ambulatory devices were

7 Ambulatory Metabolic Analysis 26 Figure 3 - The relationship of air flow volume measured with a calibration syringe and the Aerosport KB1-C using the low flow (A), the medium flow (B), and the high flow pneumotachometer (C), relative to the lines of identity. A C B limited by assessment of metabolic activity based on direct measurement of F E O 2 and the assumption of an RER of unity (4-7). As accurate computation of VO 2 and VCO 2 require measurement of fractional expired gases for O 2 and CO 2, the design of this study allowed us to detect whether errors would be introduced into this computation by inaccuracies in direct measurement of expired gases. The O 2 analyzer used in the KB1-C is a galvanic fuel cell like that used in the MMC. Measurement of F E O 2 showed a strong correlation between the analyzers, with an R 2 of Direct comparisons showed significantly lower values when measured by the KB1-C, although the magnitude of the difference was small and its clinical significance questionable. The R 2 for F E CO 2 was 0.89, with the KB1-C also producing slightly lower values of FECO 2 than the MMC. It is important to note that the statistical significance of the relatively small differences between the MMC and the KB1-C assessments of F E O 2 and F E CO 2 are a function of the large amount of statistical power available as a result of a large sample size. The mean differences between the two means of measurement were 0.26% and 0.28% for F E O 2 and F E CO 2, respectively, which are minimal effect sizes when one considers the F E O 2 ranged during the testing from 14 to 19% and the F E CO 2

8 Ambulatory Metabolic Analysis 27 ranged from 2.5 to 6.5%. It is also reasonable to expect that small differences in measurement might be observed between the two units, as they draw their gas sample by different methods. The KB1-C draws a micro sample of expired gas passing through the pneumotachometer based upon pressure differences sensed across its fixed plate. The sample for each expired breath is drawn into the gas sensors, analyzed, and then averaged over a programmed sampling period of 20 seconds to one minute. By contrast, expired gasses entering the Horizon system undergo mixing in a three liter chamber, in this case slightly downstream from the pneumotachometer, from which a single sample is drawn and then analyzed. Thus, differences in methods of analysis due to the time of sampling may explain the small differences in the F E O 2 and F E CO 2 measurements. Unlike previous reports, this study compared air flow rates measured by the three system pneumotachometers with a calibrated standard. Each of these pneumotachometers has a recommended flow range. When steady state measurements within the recommended flow ranges are taken, it is more likely that measurement of flow is accurate. However, when graded work is performed, expired air volumes may approach or exceed the end-ranges for which they are rated. The pneumotachometer setting cannot be adjusted once exercise testing has been initiated. As such, the choice of pneumotachometer for the KB1-C becomes an important component in accurate measurement of exercise metabolism. For this reason, the three pneumotachometer settings were examined beyond the recommended ranges. No other study has undertaken such an examination which delimits the air flow ranges in which this analyzer can be used. For the low flow pneumotachometer rated from 2-30 L/min, errors were small for flow rates between 9-45 L/min. However, pneumotachometer flow measurements deviated from the calibration syringe by 0.37 L/min (12.3%) at 3 L/min and 0.93 L/min (15.5 %) at 6 L/min. These results indicate that the low flow pneumotachometer setting would best be used for resting studies, resting values taken before low intensity work, or measurement of peak work in highly deconditioned persons. Caution should be exercised when interpreting data derived with the low flow pneumotachometer in conditioned persons at moderate or high levels of exertion. For the medium flow head, which is rated for flow between L/min, error was less than 3.0% percent when measuring flow rates 6 L/min to 108 L/min, and 4.6% at 117 L/min. Interestingly, the flow measurements had less error for the medium than low pneumotachometer setting at all flow rates between 6 and 63 L/min. For the high flow setting, errors at or less than 3.3% were observed for flow rates ranging from L/min. However, from 27 to 72 L/min, errors of 3.9 to 8.8% were observed. The design of this study did not allow comparison of VO 2 simultaneously derived with the KB1-C and a laboratory standard. Open bi-directional flow is required for accurate air volume assessment with the KB1-C pneumotachometers. Therefore, assessment of V E with the KB1-C system in series with a unidirectional system of descending pathway diameter is rendered suspect. It is acknowledged that the assessment of VO 2 as measure of energy uptake is a primary utilization of metabolic analyzers. Future studies are recommended to examine the accuracy of VO 2, VCO 2, and RER as assessed with this ambulatory metabolic analysis system. CONCLUSIONS This study found that fractional gas measurements were similar when assayed by an ambulatory metabolic analyzer and a laboratory standard. Accurate measurement of air flow depended upon selection of a pneumotachometer that could accommodate the peak ventilation input. The MFR satisfied this need best, and was consistently accurate at flow rates between L/min. Therefore, this pneumotachometer setting is

9 Ambulatory Metabolic Analysis 28 appropriate for most activity assessments, but would exclude accurate metabolic measurement during graded incremental testing of conditioned persons who have higher maximal ventilation values. While use of the HFP may satisfy this need at maximal work, its use will sacrifice accuracy of some data collected at submaximal work intensities. Otherwise, the physical characteristics of the unit make it practical for field testing of human performance. REFERENCES 1. Melanson EL, Freedson PS, Hendelman D, Debold E (1996) Reliability and validity of a portable metabolic measurement system. Can J Appl Physiol 21(2): Novitsky S, Segal KR, Chartr-Aryamontri B, Guvakov D, Katch VL (1995) Validity of a new portable indirect calorimeter: the Aerosport TEEM 100. Eur J Appl Physiol 70: Wideman L, Stoudemire NM, Pass KA, McGinnes CL, Gaesser GA, Weltman A (1996) Assessment of the Aerosport TEEM 100 portable metabolic measurement system. Med Sci Sports Exerc 28(4): American College of Sports Medicine (1995) ACSM s Resource Guidelines for Exercise Testing and Prescription. Baltimore, MD: Williams and Wilkins; Bruce RA, Kusumi F, Hosmer D (1973) Maximal oxygen intake and nomographic assessment of functional aerobic impairment in cardiovascular disease. Am Heart J 85: Versteeg PGA, Kippersluis GJ. (1989) Automated systems for measurement of oxygen uptake during exercise testing. Int J Sports Med 10: Aerosport Inc. (1995) KB1-C: Ambulatory Metabolic Measurement System (1995) Ann Arbor, MI:Aerosport Inc. Address Correspondence to: Patrick L. Jacobs, Ph.D., The Miami Project to Cure Paralysis, University of Miami School of Medicine, 1600 Northwest 10th Avenue, R-48, Miami, Florida 33136; Phone: ; Fax: PJACOBS@miamiproj.med.miami.edu 4. Crandall DG, Taylor SL, Raven PB (1994) Evaluation of the Cosmed K2 portable telemetric oxygen uptake analyzer. Med Sci Sports Exerc 26(1): Lothian F, Farrally MR, Mahoney C (1993) Validity and reliability of the Cosmed K2 to measure oxygen uptake. Can J Appl Physiol 18(2): Lucia A, Fleck SJ, Gotshall RW, Kearney JT (1993) Validity and reliability of the Cosmed K2 instrument. Int J Sports Med 14(7): Peel C, Utsey C (1993) Oxygen consumption using the K2 telemetry system and a metabolic cart. Med Sci Sports Exerc 25(3):